R A D I O L O G I CC L I N I C S

O F N O R T H A M E R I C A

Radiol Clin N Am 44 (2006) 41–62

41

Modern Emergent Stroke Imaging:Pearls, Protocols, and PitfallsMark E. Mullins, MD, PhD

a,b,*,1

& Background Signs and pitfalls

EpidemiologyMotivations for stroke imagingTherapeutic windowSnapshot of the recent history of strokeimaging

& Goals for stroke imaging& Radiology triage of the stroke patient& Protocols

Scenario classification and where to startRenal failureAccessContrast allergyHyperacute-acute strokeSubacute strokeChronic strokeNeck CT angiography without noncontrasthead CT

& InterpretationPearls

a Division of Neuroradiology, Massachusetts General Hob Department of Radiology, Harvard Medical School, Bo* Division of Neuroradiology, Massachusetts General HoE-mail address: Mark.Mullins@emoryhealthcare.org.1 Present address: Department of Radiology, 1364 Clifto

0033-8389/06/$ – see front matter © 2005 Elsevier Inc. All rightsradiologic.theclinics.com

Infarct evaluation on angiography(CT angiography and catheterangiography)

Factors to evaluate on CT angiography ofthe head and neck for stroke

Perfusion imagingAssessment of the findings: synthesis andputative etiologies

Stroke follow-up imaging& What imaging examination is really the

best?Detection of blood productsDetection of infarctAccess and time of examinations

& Summary& References

Stroke remains one of the most important clinicaldiagnoses for which patients are referred to theradiologist for emergent imaging. Timely and accu-rate imaging guides admission from the emergencydepartment or transfer to a hospital with a dedi-cated stroke service, triage to the intensive care unit(ICU), anticoagulation, thrombolysis, and manyother forms of treatment and management. It isimportant to approach each patient’s imagingneeds logically and tailor each work-up. Moreover,it is important constantly to review the entire pro-cess for potential improvements. Time saved in

getting an accurate diagnosis of stroke may indeeddecrease morbidity and mortality. This article dis-cusses the current management of stroke imagingand reviews the relevant literature.

Background

Epidemiology

Stroke is the third most common cause of death inthe United States, approximating 7% and trailingonly heart disease and cancer [1]. In recent years, it

spital, Boston, MA, USAston, MA, USAspital, 55 Fruit Street, GRB 285, Boston, MA 02114.

n Road NE, Atlanta, GA 30322.

reserved. doi:10.1016/j.rcl.2005.08.002

42 Mullins

has become evident that estimates of stroke inci-dence were either rising or underestimated [2].Traditionally, the rate of stroke in the UnitedStates was thought to be 500,000 per year basedon studies of patients in Rochester, Minnesota,and Framingham, Massachusetts, which were thenextrapolated to larger populations. A study by Bro-derick and colleagues [2] in 1998 using patients inCincinnati, Ohio, and within adjacent Kentuckyindicated the incidence of stroke was more like700,000 per year. Williams’ data [3] confirmedthis estimate, identifying 712,000 strokes in 1995that increased to 783,000 in 1996. This calculationrepresented an overall rate for occurrence of totalstroke (both initial and recurrent) to be 269 per100,000 population per year [3]. Williams [3]hypothesized that the total increase was caused byboth increasing age of the population and thepopulation gain.

Motivations for stroke imaging

Stroke is a common disease and any radiologistinvolved in neuroimaging is also involved in theimaging diagnosis of stroke [4]. Not only does thedisease process carry a high mortality, but also vari-able (but typically high) morbidity. Stroke is theleading cause of severe disability in the UnitedStates and the leading diagnosis for disposition ofpatients from hospitals to long-term care facilities[1]. Treatment does exist, however, for many pa-tients [5]. There is a potential for prominent impactof radiology imaging on patient care [6].

Therapeutic window

The early diagnosis of ischemic stroke is critical tothe success of therapeutic interventions, such asthrombolysis and anticoagulation. Prior studieshave indicated the time-critical nature of this dis-ease, with only a narrow therapeutic window in thefirst few hours following stroke ictus, and a dra-matic rise in hemorrhage complications thereafter[7–20]. Diagnosis in the first 3 hours postictusprovides the opportunity for intravenous or intra-arterial thrombolysis and intra-arterial clot me-chanical treatment (attempts at removing the clotor breaking it into smaller pieces), which has beenshown to improve outcome [12,15,16,21]. Diagno-sis in the time period between 3 and 6 hoursprovides an opportunity for intra-arterial thrombo-lysis and mechanical treatment. Diagnosis in thefirst 12 hours provides the opportunity for admin-istration of neuroprotective agents, which mayimprove outcome. Involvement of the posteriorcirculation, especially the basilar artery, is treatedby some physicians regardless of time of onset orup until 12 to 24 hours in some practices. This isrelated to the potentially high mortality and mor-

bidity associated with basilar artery thrombosis.New studies are being performed to evaluate thepossibility of basing treatment on imaging ratherthan a time window related to ictal onset, but theseprotocols remain strictly experimental at the timeof this writing.

Snapshot of the recent history of strokeimaging

The mainstays of early stroke diagnosis are con-ventional noncontrast head CT (NCCT) and con-ventional brain MR imaging [10,19,22–28]. Somefacilities routinely use MR imaging with diffusion-weighted imaging (DWI) in the detection of acutestroke [29] because it has been shown to yield im-proved sensitivity, negative predictive value, andaccuracy compared with NCCT and conventionalMR imaging [17,21]. CT angiography (CTA) of thecerebral vasculature has been promoted by someauthors as a means to diagnose the vascular occlu-sions of acute stroke [13,14,27,30–32], especiallyin the hyperacute (<3 hours postictus) setting whendecisions regarding intervention are paramount[11,33]. Specifically, CTA can reliably identify fill-ing defects within the circle of Willis or its proximaltributaries (most commonly the middle cerebralartery [MCA] [34] and its branches) and parenchy-mal filling-perfusion defects indicative of whole-brain perfused blood volume [24]. CTA has beenshown to be very useful in the diagnosis of MCAembolic stroke and predictive of infarction volumein the MCA distribution [11,30]. In contradistinc-tion, CTA may be less useful for stroke involvingthe deep gray matter or brainstem [30].

Goals for stroke imaging

The theoretical goals for stroke imaging include

1. Access to high-quality equipment: The author’sinstitution uses frequently updated hardwareand software and places MR imaging and CTscanners in the emergency department, beingboth close to each other and to the patients andreferring physicians so that examinations maybe done in short succession.

2. Ability to perform specialized examinations: Multi-detector row CT scanners are necessary for mostmodern specialized CT examinations includingCTA (hereafter all CTA denotations imply aninitial NCCT followed by CTA unless otherwisenoted) and CT perfusion (CTP). The author’s in-stitution has a dedicated three-dimensional (3-D)laboratory with technical coverage 24 hours aday to provide maximum intensity projections,volume-rendered reconstructions, and perfu-sion maps. Multiplanar reformations are avail-

Box 1: Ictal symptom onset times

1. Undefined, unknown, or unclear onset time:Traditionally these patients are not eligiblefor thrombolysis

43Modern Emergent Stroke Imaging

able automatically from many modern CTconsoles. The author also reviews images onsoft copy and recommends a patient archivingand communication system for subsequentimage manipulation and scrolling. MR im-aging including MR angiography, DWI, andperfusion-weighted imaging (PWI) is stan-dard on most modern MR imaging devicesbut does necessitate at least a 1.5-T magnetstrength, echo planar imaging modalities, anda power injector.

3. Accurate and timely diagnosis: Imaging does notbenefit the patient until interpreted; the author’sinstitution provides 24-hour coverage for con-sultation, protocoling, performance, and inter-pretation of patients’ examinations. The fourprimary diagnostic questions are as follows:a. Is there an infarct (an imaging manifesta-

tion of cytotoxic edema)?b. Is there intracranial hemorrhage? The stroke

physicians do not alter treatment based onpetechial hemorrhage but that other types ofhemorrhage probably alter treatment plans.

c. Can stroke mimics, such as encephalitis ortumor, be excluded?

d. What portion of brain is completely in-farcted and what part is salvageable or mani-fests as brain at risk of infarction? Withsome variability, decreased or restricted dif-fusion on DWI defines the former [35] andperfusion-diffusion mismatch on PWI orCTP defines the latter.

2. <3 hours postictus: Intravenous thromboly-sis candidate

3. <6 hours postictus: Intra-arterial thrombo-lysis candidate

4. >6 hours postictus: Usually the patient is nota thrombolysis candidate but this contra-indication may be made more relative inthe following scenarios:a. Basilar artery thrombus: lifesaving attempts

are made in somewhat heroic circum-stances in some of these cases, especiallyif the patient is young and otherwisehealthy

b. Ictal onset time surrogates: perfusionimaging; replacing the archetype of ictalonset time with surrogates, such as per-fusion imaging, is not yet ready for frontline use but is being studied and shouldbe considered to be controversial and atbest experimental under current stan-dards of care in terms of replacing ictal

4. Vascular imaging: Fisher [36] identified in 1951that most stroke was thromboembolic and fur-thermore was related to atherosclerotic carotiddisease of the neck. The basic philosophyyielded is to image the neck arteries when strokeis suspected (the author prefers CTA to MR an-giography or ultrasound but finds that the sur-geons and neurologists are most comfortablewhen combinations are used).

5. Radiation control: One should not lose sightof the need for reduction of radiation expo-sure to the patient [37]. Briefly, the conceptof ‘‘as low as reasonably allowable’’ shouldbe implemented and examinations should beclinically indicated.

6. Cost: Although it is difficult to prove with cer-tainty, many clinicians believe that an accurateimaging diagnosis likely decreases overall costto the system.

onset time. It is, however, of great use interms of understanding the patient’sphysiology and many believe that pa-tient management will soon be basedprimarily on these modalities, followingsuccessful studies.

Radiology triage of the stroke patient

First, a good history (not necessarily long or de-tailed, just accurate and appropriate) should beobtained. A proper history improves the ability to

make a correct diagnosis [38]. Does the patient’spresentation sound in any way like a stroke? If so,the patient has a high likelihood of stroke (perma-nent) or transient ischemic attack and imaging isindicated. If stroke is possible, expedite the neuro-imaging (be aggressive in terms of getting imagingquickly and appropriately). If the hospital has anavailable neurologist or better yet a stroke team, askthe referring physician strongly to consider callingthem immediately. Time is of the essence.Ask the referring physician, patient, or family

members about symptom (ictal) onset time. Ingeneral, the patient falls into the categories listedin Box 1.Neuroimaging directly affects the disposition

of the patient from the emergency department,whether they are discharged; admitted to the regularfloor or to (neuroscience) ICU; and under whatservice they are admitted and managed. Treatmentoptions guided by neuroimaging include anti-coagulation; thrombolysis; mannitol or steroids;blood pressure control (hypertension or hypoten-sion); hypervolemia-hemodilution; craniotomy; andplacement of an intracranial pressure-measuring de-vice. For patients presenting in an acceptable time

44 Mullins

frame for thrombolysis, the first task of neuro-imaging (usually NCCT in most centers) is to dem-onstrate lack of intracranial hemorrhage and lackof large territorial infarct (usually MCA distribu-tion) [39–42]. Contraindications for thrombolysisinclude but are not limited to recent head surgery,gastrointestinal bleeding, and other bleeding dia-theses. For additional and more specific indicationsfor thrombolysis with intravenous or intra-arterialmedications, the reader is directed to the reviewsin the reference section. All of these factors should

Fig. 1. A 49-year-old woman with transient ischemic attaterritorial infarct is illustrated with a right insular ribbon sigand (B) infarct extent better visualized using stroke windoalso manifest on the CTA source images with strokewindowdefect on a CTA maximum intensity projection (MIP) imainfarct hyperintensity (long arrows) and prominent corticawith acute infarct age. Axial DWI illustrates hyperintensityarea abnormal on CTA (arrows), consistent with acute infararterial input functions of the left internal carotid artery (ICarrows) (the author does not primarily use the contralatabnormality but does recognize that this is controversiadecreased cerebral blood flow (K) that partially normalizwith mean transit timemaps (short arrow, J). This constellatinfarct and a larger territory at risk for infarction, and s

be evaluated while the patient is getting queuedup for scanning.

Protocols

Scenario classification and where to start

If the diagnosis is unknown and the scenario is non-acute or late acute, work-up typically begins withNCCT, with decisions for additional imaging basedon the results. Hyperacute is generally meant to ex-

ck and headache and visual field loss. (A) Right MCAn on noncontrast head CT with brain windows (arrow)ws (arrows, border infarct extent). (C ) Infarct extent iss (arrows). (D, E ) Right MCA clot is visualized as a fillingge (arrows). (F ) Axial FLAIR imaging illustrates partiall vascular hyperintensity (short arrow), most consistent(G) and ADC hypointensity (H) within a portion of thect. (I) There is increased mean transit time on PWI usingA) (bordered by arrows) and (J) right ICA (bordered byeral input function because it overestimates the truel so both are performed). There is a similar area ofes on the cerebral blood volume map (L), correlatingion of findings is most consistent with amoderate-sizeduggestion of some, but incomplete collateralization.

45Modern Emergent Stroke Imaging

press symptoms lasting fewer than 3 to 6 up until12 hours, whereas ‘‘acute’’ may mean a few hoursto several days, but usually means <24 hours. ‘‘Sub-acute’’ generally means several days to weeksand ‘‘chronic’’ means months to years (typically>3 months). The literature regarding terminology isvariable, as are local usage customs. The best goalis to be descriptive, specific, and internally con-

Fig. 1 (continued ).

sistent. The specific goal should be effectively tocommunicate the impression of the case to the re-ferring physician.

Renal failure

If there is known or suspected renal failure in apatient not currently receiving routine dialysis, thentriage of the patient to MR imaging without con-

Fig. 1 (continued ).

46 Mullins

trast material (and without PWI) is appropriate ifno other MR imaging contraindications exist. If thepatient is on dialysis, then dialysis should follow inshort succession any CT scan or MR image withcontrast. Renal failure does not necessarily createan indication to use gadolinium-based contrastmaterial because there are theoretical risks ofheavy metal poisoning if the chelate disassociates(refer to the particular package insert for details). Ifthe patient is to receive catheter angiography, theCTA may be obviated. The author uses creatininevalues of <1.5 for normal, 1.5 to 2 for borderline(relative contraindications), and >2 for contraindi-cation to contrast material without dialysis. Inemergent circumstances, the patient’s physicianmay approve the use of contrast material withcreatinine value still pending or within the borderzone region; in these cases, the risk for transientand permanent renal failure is increased. A renalmedicine consult may be obtained if time allows.Use of standard medications and therapies, suchas hydration and N-acetylcysteine, is no differentthan with any other patients receiving iodinatedcontrast material.

Access

MR imaging PWI, CTA, and contrast-enhanced MRangiography require adequate peripheral intrave-nous access, typically larger than 20 gauge. Larger-access cannulas are necessary for higher flow ratesand access should be optimized to effect technicallyadequate results. Flow rates may be adjusted downfor smaller access and known atrial fibrillation-arrhythmia or decreased cardiac output but the au-thor’s experience with bolus-chasing CT and MRimaging techniques has been anecdotally variable.Instead, in most instances the author simply ad-justs the imaging time start point to 10 seconds

later in patients with these known diagnoses andthis is sufficient to obtain interpretable images inmost circumstances.

Contrast allergy

Iodinated contrast material allergy is an indicationfor MR imaging or MR angiography, or gadolinium-based contrast-enhanced CTA or catheter angiog-raphy. Images obtained with the latter two aretypically diagnostic [43].

Hyperacute-acute stroke

For patients with hyperacute or acute stroke, theauthor uses CTA of the head and neck combinedwith CTP, followed by conventional brain MR im-aging including DWI and susceptibility sequences[Figs. 1–3]. This version of CTA begins scanning atthe skull base, extends up to the vertex, and thenstarts again at the arch of the aorta to completeimaging of the neck vessels. It is usually of greatvalue to include the great artery and vertebral arteryorigins. Repeating arterial imaging with MR angi-ography after CTA is usually not indicated unlessthere is a problem-solving aspect to the MR angi-ography. T1 fat-saturated sequences for potentialarterial dissection should be considered when set-ting up the MR imaging. If there is no territory atrisk of infarction on perfusion imaging (no penum-bra), it is thought that the patient is at increasedrisk of hemorrhage with thrombolysis; the risks ofthrombolysis and mechanical intra-arterial therapylikely outweigh the benefits [Fig. 4]. Initial imagingshould address this question.Several different combinations of CT, MR imaging,

perfusion imaging, and vascular imaging can beused to evaluate acute stroke. The author choosesinitial noncontrast head CT followed immediately

47Modern Emergent Stroke Imaging

by CTA and CTP to provide a rapid yet thoroughassessment of potential intracranial hemorrhage;stroke mimics; large evolved infarct (contraindica-tion to thrombolysis); arterial clot and stenosis;infarct size and location; and penumbra. MR imag-ing adds primarily DWI, which frequently improvesdetail as to the presence and extent of infarct.Many practitioners do not have quick access to

MR imaging for their stroke patients and CT im-aging dominates those diagnostic work-ups. Quickaccess to MR imaging suggests that MR imagingmay also be used in lieu of CT scanning but inmost situations CT scanning can be performed evenbefore the patient is cleared and prepared for MRimaging. In some instances, more useful data canbe obtained with CT methods; for example, theauthor finds that the increased detail in vessel im-aging obtained with CT angiography is frequentlymore helpful than MR angiography. Because manyradiologists and referring physicians believe that CTis superior to MR imaging for the assessment ofintracranial hemorrhage, first-line MR imaging (inlieu of CT scanning first) remains nonstandard. Infact, MR susceptibility imaging is clearly superior toCT in identifying hemosiderin blood products thatmay increase the risk of intracranial hemorrhageduring stroke treatment. Some data presentedwithin the past several years have suggested thatMR imaging at least as good at detecting hemor-rhage, but a definitive study is indicated to convertcurrent practice patterns. Finally, it must be re-emphasized that there is no single absolute proto-col to follow and the process should be constantlyre-examined for potential improvements.

Subacute stroke

For patients with subacute stroke, the author usesCTA of the head and neck without perfusion, fol-lowed by conventional brain MR imaging includ-ing DWI and susceptibility. CTP is generally notperformed because most available treatments forpatients in this clinical scenario do not involve per-fusion data input (eg, carotid endarterectomy).There may be situations, however, in which CTPor PWI is appropriate for an individual patient.

Chronic stroke

For patients with chronic stroke the author usesconventional brain MR imaging including DWIand susceptibility combined with MR angiogra-phy of the head and neck. The author routinelyperforms neck MR angiography with gadolinium-based contrast material and MR angiography of thecircle of Willis using 3-D time-of-flight methodol-ogy (unenhanced). Two-dimensional time-of-flight

MR angiography of the neck may complement 3-Dcontrast-enhanced MR angiography in some pa-tients. Phase-contrast imaging is useful for determi-nation of arterial flow directionality.

Neck CT angiography without noncontrasthead CT

Neck CTA only (both without initial NCCT andwith only a portion of the circle of Willis imaged)may be performed but the author does not recom-mend this option unless this is to be used forproblem solving. For example, clarification of anabnormal ultrasound result is indicated.

Interpretation

Pearls

The traditional reason for performing NCCT ini-tially is to exclude hemorrhage and obvious non-infarct disorders. Although useful in this regard,NCCT may also yield signs leading directly to adiagnosis of infarct.In a recent study of patients, hemorrhage on NCCT

was identified in approximately 5% [44]. This bloodis thought to represent primarily hemorrhagic trans-formation of ischemic infarct. In younger patients,however, underlying vascular lesions should beconsidered. If a nonarterial-distribution abnormal-ity with hemorrhage is identified, consider venousinfarct. Delayed hemorrhagic conversion is mostlikely related to late increased arterial flow or col-laterals to damaged brain (ie, reperfusion injury)or coagulopathy.When proctoring a CTA, account for patency of

the bilateral internal carotid and vertebral arteries.If there is lack of apparent contrast material fillingon the initial images, the author immediately per-forms delayed images to see if there is delayed fill-ing as a manifestation of hairline lumen (whichusually is treated surgically) as opposed to occlu-sion (which likely does not undergo surgery) [45].Anecdotally, this is at least somewhat clinicallyuseful in every case in which it is performed.When proctoring a CTP, consider splitting the

entire contrast material bolus to cover a largerarea of brain (ie, two or more scans) and be sureto include an artery in the area imaged (slab) thatcan be used to perform the technical aspects ofthe map performance (ie, arterial input function).There is controversy regarding whether to use theipsilateral or contralateral arteries for an arterialinput function and the author has chosen to usethe ipsilateral artery because the contralateral arterygives an overestimate of perfusion abnormality.

48 Mullins

If a susceptibility-weighted sequence is not avail-able or not obtained, the author has found anec-dotally that similar data may be obtained from theDWI or a subset map of apparent diffusion coeffi-cient called low-B (setting the B value to nearly 0 forpreparation of the map). Because the contrast onthe baseline acquisition from diffusion-weightedsequences is T2 and not T2* (as in the dedicatedsusceptibility) sequence, it is likely that the DWIdata are not optimal for detection of magneticsusceptibility artifact related to blood products;furthermore, its use has not been validated. The

Fig. 2. A 16-year-old man with hemiparesis and neck pain.right lentiform nucleus sign on noncontrast head CT withFLAIR imaging illustrates partial infarct hyperintensity (long(short arrow), most consistent with acute infarct age. (D) Ax(arrow) and lentiform nucleus hyperintensity, most consisteAn axial gradient echo susceptibility sequence image (E) il(F ) Postcontrast axial T1-weighted imaging illustrates intrahyperintensity (arrow) and (H) ADC hypointensity (arrow) wacute infarct. (I) Loss of flow-related enhancement is notearrows) on 3-D time-of-flight MR angiography of the circleincreased mean transit time on PWI using both in the infaand within the posterior division territory of the right MCDWI. There is an essentially matched area (arrows) of dnormalizes on the cerebral blood volume map (L). Thismoderate-sized completed infarct and a larger territory a

dedicated magnetic susceptibility (T2*) sequenceis preferable [46,47].

Signs and pitfalls

Acute infarct hallmarks on routine imaging [48]include obscuration of gray-white matter interfacerelated to cytotoxic edema; wedge-shape; vascu-lar territorial; decreased tissue enhancement; andrestricted diffusion (DWI hyperintense, apparentdiffusion coefficient (ADC) hypointense, and expo-nential map hyperintense [49]). The intravascu-lar enhancement sign is characteristically visible

Right basal ganglia infarct (arrows) is illustrated with abrain windows (A) and stroke windows (B). (C ) Axialarrow) and prominent cortical vascular hyperintensityial T2-weighted imaging redemonstrates right caudatent with an infarct of several hours to a few days in age.lustrates no evidence for hemorrhagic transformation.vascular enhancement (arrow). (G) Axial DWI illustratesithin the area of abnormality on CTA, consistent withd within a portion of the right M1segment (betweenof Willis, consistent with nonocclusive clot. (J) There isrcted area defined by restricted diffusion (long arrow),A (between short arrows), not yet infarcted based onecreased cerebral blood flow (K) that only minimallyconstellation of findings is most consistent with a

t risk for infarction, with incomplete collateralization.

49Modern Emergent Stroke Imaging

within the first 3 days (75%), best seen on spinecho sequences and remaining one of the earliestand most sensitive signs of acute infarct [50]. Thissign is relatively nonspecific, however, and resolveswithin 5 to 7 days, providing a narrow windowof detection. The dural (meningeal) enhancementsign is best seen at the tentorium, and on coronalimaging [50]. This sign appears within the first3 days and resolves by 7 days in most patients.Gyral enhancement of infarct is best recalled ac-

Fig. 2 (continued ).

cording to Elster’s rule of 3’s: as early as 3 days,maximal at 3 days to 3 weeks, and gone by3 months [50]. Use of double- or triple-dose gado-linium contrast material or use of single-dose gado-linium contrast material with magnetic transfertechniques effects earlier appearance (1 day) anda longer-lasting sign (up to 6 months) [50], al-though such an approach is rarely necessary.On MR imaging, hyperintensity on T2-weighted

and fluid-attenuated inversion-recovery (FLAIR)

50 Mullins

images takes several hours to a few days to develop.This observation can be useful to age an infarct.Phase and evolution hallmarks include the follow-ing milestones (caution is advised because theseguidelines vary):

1. Hyperacute infarct (< 6–12 hours): NCCT islikely normal and DWI is abnormal in approxi-mately < 30–120 minutes and abnormality per-sists up to 2 weeks [35,50].

2. Acute infarct (12–48 hours): Meningeal en-hancement, edema, and mass effect. Fogging

Fig. 3. A 67-year-old man with weakness. (A) A small, distalMIP reconstruction from CTA source images. (B) In the nlumen (long arrow) with atherosclerotic plaque (short arrowlarge area of prolonged mean transit time within the ledecreased cerebral blood flow (D, arrows), and nearly normost consistent with territory at risk for infarction, andimaging without thrombolysis (the patient did not meet inillustrates acute left MCA infarct on axial FLAIR (F ) parhyperintensity (short arrow), no abnormal enhancement ohemorrhagic transformation on susceptibility-weighted imDWI (H) and decreased on ADC maps (I). PWI illustratescerebral blood flow (arrows) and (L) cerebral blood volumecompleted infarct, which was proved on follow-up imaginchanged between the scans most consistent with propagathrombolysis inclusion criteria will depend more on surrog

and pseudonormalization may occur on NCCTand ADC-DWI [50].

3. Subacute infarct (2 days–2 weeks): Parenchymalenhancement commences at 4 days; edema re-solves (maximal at 3 days); and mass effectdecreases at 7 to 10 days [50]. Luxury perfusionis most evident and is likely best evidenced byincreased cerebral blood volume (CBV) on PWI.In the author’s experience, CBV is the mostaccurate initial parameter to predict outcomeand final infarct volume but is not routinelyobtained or necessary in the subacute setting.

left MCA nonocclusive clot (arrow) is demonstrated oneck, the CTA MIP image illustrates a hairline residual) within the proximal left cervical ICA. CTP illustrates aft MCA distribution (C, arrows), essentially matchedmalized cerebral blood volume (E); these CTP data aregood collateralization. Immediately subsequent MR

clusion criteria in that the ictal onset time was unclear)enchymal (between long arrows) and cortical vesseln T1-weighted imaging (not shown), no evidence forage (G), ill-defined and variably increased signal on(J) increased MTT (arrow) and (K) relatively matchedmaps (arrows); this appearance is most consistent withg (not shown). The implications of the perfusion datation of clot. Perhaps in the future, decisions regardingate markers, such as imaging results.

51Modern Emergent Stroke Imaging

4. Chronic infarct (> 2 weeks): Encephalomalacia(appears at 6–8 weeks) and cortical laminarnecrosis [50].

Infarct signs on NCCT [51] include the following:insular ribbon sign; obscuration of the lentiformnucleus; hyperdense artery; and the MCA dot (enface) and dash sign (in profile) [52]. The author’s

Fig. 3 (continued).

data [52] suggest that not all dense arteries corre-spond to clot on CTA. Despite potential biasesof CTA, it is nonetheless used for acute decision-making and the correlation to actual clot is thoughtto be excellent [14].Any portion of the stroke imaging process is

subject to potential pitfalls, and a few are describedhere. During acquisition of CTA source images it is

Fig. 3 (continued ).

52 Mullins

preferable to review the images quickly while thepatient is still on the scanner for at least a fewreasons: (1) assessment likely governs what treat-ment the patient receives (especially anticoagulantsand thrombolytics) and any time saved may impactthe results; (2) the contrast material could haveextravasated and it is important to note this andtreat the patient appropriately; and (3) if the typi-cal important four vessels (bilateral internal ca-rotid and vertebral arteries) within the neck arenot opacified on all images, then delayed imagesmay be obtained to assess whether the finding rep-resents complete occlusion (typically not treatedsurgically or endovascularly) or is a critical stenosisor hairline residual lumen (treated surgically inmost cases). If this is not recognized initially, ad-ditional imaging, such as sonography, catheterangiography, or MR angiography, needs to be con-templated. Another pitfall pertains to the identifi-

Fig. 4. A 60-year-old woman with hemiparesis. Intra-cranial hemorrhage is illustrated (hyperdensity) onnoncontrast head CT following intra-arterial tissueplasminogen activator administered for left MCA ter-ritorial infarct (arrows border infarct).

cation of artifacts on CT scans and MR imaging,particularly DWI.DWI used routinely (at 1.5-T field strength) may

not identify all infarcts (false-negatives), mostcommonly in the posterior fossa, brainstem, andalong areas near the sinuses and surface of thebrain. Moreover, DWI may indicate abnormalitiesthat eventually resolve (false-positive for infarct).These instances are sometimes related to ischemiawith or without treatment but may also be seenwith seizure activity and demyelination. Brain tu-mors may show restricted diffusion and have occa-sionally been mistaken for infarcts, at least initially.It is important to use all facets of the MR imagecharacteristics to determine whether the lesion fitstypical criteria across all sequences (eg, subacuteinfarct with T2 hyperintensity, DWI hyperintensity,ADC isointensity, possible gyral enhancement, andno mass effect). If the lesion does not demonstratean expected appearance across all sequences, ashort-term follow-up examination may be indi-cated to confirm expected evolution. Correlationto the clinical scenario may also benefit the diag-nosis because stroke is an acute event, whereastumor is typically insidious in onset.A search for motion artifact on perfusion imag-

ing should be evaluated in every case performedbecause the resultant perfusion maps may be erro-neous and misleading, or simply uninterpretable.Evaluation of the source images using cine modeon a softcopy review system is usually helpful forevaluating the initial data before map formationand to evaluate for motion.

Infarct evaluation on angiography(CT angiography and catheterangiography)

CTA is an excellent first-line examination for evalua-tion of the head and neck arteries [Figs. 5 and 6]

Fig. 5. A 78-year-old man with hemiparesis. Left MCA territorial infarct is illustrated with (A) distal left MCA fillingdefect consistent with thrombus (arrow), (B) increased mean transit time on CTP (bordered by arrows), and (C ) asmaller area of decreased cerebral blood flow that (D) partially normalizes on the cerebral blood volume map.This constellation of findings is most consistent with a small infarct and a larger territory at risk for infarction, andsuggestion of good collateralization. (E) Follow-up noncontrast head CT confirms this assessment because the finalinfarct volume (between arrows) approximates the area of infarct only on CTP, following appropriate stroketreatment (including thrombolysis).

53Modern Emergent Stroke Imaging

Fig. 6. A 61-year-old woman with hemiparesis. Nonocclusive left MCA thrombus on 3-D imaging performed fromCTA source data (A) and was confirmed at catheter angiography (B), illustrating the typical accuracy of CTA foridentification of thrombus. The arrows point to the proximal aspect of the thrombus.

54 Mullins

[14,53]. In comparison with catheter angiography,what it gives up in resolution and dynamic prop-erties it makes up for with rapid acquisition of a3-D dataset and the potential for assessment ofwhole-brain perfusion [54] and better parenchymalevaluation [14]. Moreover, CTA illustrates not onlyarterial stenosis or occlusion, but also the vesselwall. This factor is most important for evaluationof intramural dissection [Fig. 7] and thrombosedaneurysms, which may both complicate evaluationof stroke patients.CTA can be treated as a noninvasive angiogram,

evaluating the vessels just as would be done witha conventional angiogram. More specifically, the

Fig. 7. A 37-year-old woman with neck pain and nonfodissections are illustrated on curved reformatted image frdissection (arrow) with smoothly tapering wall, and T1 hypnoncontrast fat-saturated imaging (B). The left ICA abnorare consistent with acute or subacute right ICA dissectioimages depends on age of blood products, so the left ICAchronic dissection.

signs that are evaluated for both examinationsinclude vessel occlusion or cutoff related to throm-boemboli; aneurysms; arterial dissection; meniscusor flattened shape to clot (recent) versus reversemeniscus (older); tram track (nonocclusive or re-canalized clot); delayed (antegrade) flow (manifestas decreased whole-brain perfusion on CTA); and(retrograde) collateral flow [55].Use of stroke window and level settings has been

shown to improve infarct detection on NCCT[Fig. 8] [56]. The author routinely uses these win-dow and level settings also to evaluate CTA sourceimages, as a supplement to routine window andlevel settings.

cal clinical findings. Bilateral internal carotid arteryom CTA source data (A) illustrating a left cervical ICAerintensity within the wall of the right ICA on axial T1mality is isointense on T1-weighted imaging. Findingsn (arrow ; methemoglobin). Intensity on T1-weightedfindings (isointense) could suggest either very early or

Fig. 8. A 72-year-old woman with hemiparesis. (A) Late acute left MCA infarct is illustrated on noncontrast head CTwith brain windows (arrows border infarct); (B) noncontrast head CT with stroke windows (arrows border infarct);and (C ) 3-D imaging performed from CTA source data showing a T-lesion of clot within the left ICA, M1, and A1segments (arrow). The left ACA territory appears spared, suggesting that the primary arterial supply of the leftACA is not by the left A1 segment. (D and E) CTA 3-D reformatted images of the neck illustrate a severe proximalleft cervical ICA stenosis (arrow) that includes the origin and is relatively smooth, potentially suggesting arterialdissection but more likely atherosclerotic in a patient of this age.

55Modern Emergent Stroke Imaging

56 Mullins

Factors to evaluate on CT angiography of thehead and neck for stroke

Internal carotid artery stenosis [see Fig. 8] tradi-tionally has been described by percent stenosis be-cause of conventions used in the NASCET trials.The author, however, uses residual lumenal diame-ter because it more appropriately describes thephysiology regardless of anatomic variations inbackground artery size [45]. The author describesstenoses with measurements of the smallest re-sidual lumen and a subjective grade (mild, mod-erate, severe, critical). In the author’s facility,1.75-mm residual lumen corresponds to a 70%stenosis, but this may vary from laboratory tolaboratory. Moreover, ultrasound velocity cutoffsfor 70% stenosis vary in a similar fashion andeach group must set its own standards. Previouswork by the NASCET trial suggests a benefit ofcarotid endartarectomy for patients with symptom-atic internal carotid artery stenosis >70% [57].Identification of ulcerated plaque is also relevant

because of an increased risk of thromboembolism.This diagnosis may be made to some extent oncatheter angiography, but the author also routinelymakes this assessment on CTA. Reporting of loca-tion of calcifications or plaque without stenosisrelated to atherosclerosis may benefit the patient’sprimary caregiver using medical management. Inthe future, it is likely that identification of so-calledvulnerable plaque that is at increased risk of being asource of thromboembolic disease may be per-formed with such techniques as sonography, CTA,or MR angiography. This assessment, however, isnot used routinely at present. Hemodynamic sig-nificance on CTA, may be inferred by observingpoststenotic dilatation, and this observation isworth at least suggesting when communicatingfindings and their implications; however, it mustbe noted that this type of implication has not beendefinitely correlated in the literature and is strictlyanecdotal. In the author’s hospital, sonographyresults with velocity measurements are generallyused in conjunction with CTA results to guide sur-gical planning. Furthermore, the vascular surgeonsask for performance and agreement between resultsof two of the following three examinations beforeconsideration for endartarectomy: (1) sonography,(2) CTA, and (3) MR angiography.

Perfusion imaging

The primary goal of CTP and PWI is to determinepotential brain (territory) at risk of infarction,which is thought to represent an ischemic penum-bra that is salvageable if treated appropriately[58–63]. Several different types of maps may becomputed from the raw data and a detailed dis-

cussion of the techniques and variations in thesetechniques is beyond the scope of this article, butthe reader is encouraged to pursue additional clari-fication in the literature [64].Standardized and semiautomated software is

commercially available to assist in the task of form-ing perfusion maps. There are, however, a couple oftechnical points that should be made. First, motionartifact may not only cause the maps to appearuninterpretable but may also cause erroneous in-terpretation if this artifact is not identified. Second,there exists a controversy as to which side shouldbe used as the arterial input function for the perfu-sion characterization. The author primarily uses theipsilateral side because the data suggested that thecontralateral side overestimated the potential terri-tory at risk for infarction. Third, results may varyprominently if the user input functions are changedeven slightly [64].The routine perfusion maps that the author uses

are mean transit time(similar to time to peak),cerebral blood flow, and CBV. The author’s inter-pretation starts with the mean transit time map,because it yields typically the largest potentially ab-normal area. On this map, increased signal is bad,indicative of delayed blood supply to this brainparenchyma. On the cerebral blood flow map, de-creased signal is bad and is usually containedwithin the mean transit time abnormal region, rep-resenting delayed or decreased blood flow to thebrain parenchyma through the normal antegradearterial pathways. The CBV maps are likely the bestestimate of collateral flow. Here, decreased signalis bad, indicative of delayed or decreased bloodvolume or flux into the brain parenchyma. CBV islikely the best predictor of final infarct volume butmost patients end up with a final infarct volumesomewhere between the size of the cerebral bloodflow abnormality and the CBV abnormality (whichis usually smaller than and contained within thecerebral blood flow abnormality). Increased signalmay be obtained with luxury perfusion and reper-fusion. In practice, relative or semiquantitativeparametric maps are created for MR PWI insteadof absolute quantitation. The processing for abso-lute quantitation is not trivial, and off-the-shelfsoftware for this purpose is not widely availablefor MR PWI. There are, however, several vendor-supplied packages for quantitative CTP.CTP and PWI are currently comparable, so use

should be based on local access and practice cus-toms. If the abnormal regions on CBV and cerebralblood flow are matched, this scenario likely repre-sents completed infarct without good collaterals. Ifthe abnormal regions on cerebral blood flow aregreater than CBV, this is suggestive of some normali-zation and indicative of likely good collaterals; this

57Modern Emergent Stroke Imaging

situation is unlikely to extend to complete infarct iftreated appropriately and aggressively. This patientis likely the best candidate for therapy. If the abnor-mal regions on mean transit time are greater thancerebral blood flow or CBV, no one has shown forcertain what the area abnormal on mean transittime mean transit time only represents, but someclinicians treat this as potential territory at risk forinfarction. If there is a perfusion deficit of any kindand the patient has hypotension or hypoxia, theseregions may become ischemic or infarcted. If anarea of restricted diffusion on DWI is matched tothe perfusion abnormality, this scenario likely rep-resents a completed infarct. A caveat must be made,however, that DWI can reverse rarely and althoughsome think of restricted diffusion as infarcted tis-sue, there is controversy as to what portion of thesecases manifest ischemia. Conversely, where there isa DWI-PWI mismatch, this likely represents braintissue at risk [51] for infarction.

Assessment of the findings: synthesis andputative etiologies

Prognosis, risk of recurrence, and management op-tions are influenced by stroke subtype. Most ische-mic infarcts are thromboembolic, but the etiologyis not always clear on clinical examination. Aug-menting clinical evaluation with neuroimagingmethods, such as those described previously, canbe helpful in categorizing etiology into one of fivesubsets: (1) cardioembolic; (2) large-vessel stenotic-occlusive; (3) small-vessel occlusive; (4) other, butdetermined cause (eg, arterial dissection, vasculitis,and so forth); and (5) cryptogenic [65,66]. Visu-alizing the vessels (and often the heart and aorticarch) is a critical part of this process. In addition,

Fig. 9. A 54-year-old man with visual complaints and atriaDWI by hyperintensities within the occipital (arrows), PCAbilaterally and the left frontal (arrow) ACA territory (B).

the pattern of ischemia-infarction in the brain canprovide clues to etiology.For example, infarcts of different ages but in the

same region with a stuttering clinical course is sug-gestive of large-vessel stenosis, whereas infarcts ofdifferent ages in different arterial territories is moresuggestive of a central embolic etiology, perhapscardiac. Infarcts of the same age involving differentarterial territories suggest a central embolic source[Fig. 9], again usually related to the heart, aorticarch, or perhaps a right-to-left shunt. Multiplesmall, primarily cortical infarcts of similar ageor different ages could suggest a systemic process,such as vasculitis, but could also have a proximalembolic etiology like endocarditis. Borderzone orwatershed infarct patterns might suggest a proximalembolus or large-vessel stenosis complicated byhypotension, or a combination of these. More dif-fuse and symmetric injury patterns are seen withglobal hypoxia or anoxia, such as in the setting ofprolonged cardiac arrest. In practice, the exact etiol-ogy is not always clear, and combinations of pat-terns may be present.The combination of anterior and middle cerebral

infarcts (ACA and MCA, respectively) infarcts issuggestive of an internal carotid artery clot (alsoknown as a T-lesion [Fig. 10]), a lesion with a poorprognosis. In a patient with an embolus to theMCA, infarction of the lentiform nucleus suggestsa proximal M1 clot (also known as a stem clot) andits absence suggests clot distal to the lateral len-ticulostriate arteries. Circle of Willis variants canmodify the pattern of ischemia. For example, pres-ence of large posterior communicating arteriesmay explain involvement of anterior and poste-rior circulation territories from a single embolicevent through the internal carotid artery, and a

l fibrillation. Embolic acute infarcts are manifested onterritories bilaterally (A) and parietal, PCA territories

Fig. 10. A 59-year-old woman with hemiparesis. Similar situation to Fig. 8 but in this case the left ACA territory hasinfarcted (A), whereas the right ACA territory is spared, indicating more balanced circle of Willis configuration.T-lesion is again illustrated on the left on 3-D imaging performed from CTA source data. (B) ICA clot (short arrow);M1 clot (long arrow); A1 clot (open arrow). (C ) Left M1 clot (short arrow); left MCA calcified embolus versusatherosclerosis (long arrow).

58 Mullins

small or absent contralateral A1 segment can leadto bilateral ACA territory infarcts from a unilateralproximal embolus like an internal carotid arteryocclusion or T lesion. In general, lacunar infarctsare unlikely to be embolic, but this remains theo-retically possible. In patients with lacunar infarcts,check for signs of hypertension and atheroscleroticdisease on imaging, including tortuous arteries inthe neck and signs of leukoaraiosis (also known asnonspecific white matter change).

Stroke follow-up imaging

The following are the primary aspects to be evalu-ated on follow-up imaging of patients with stroke(in most cases, this is accomplished with NCCT[Fig. 11]):

1. Has there been extension of previous infarct?2. Is there any new location of infarct?3. Has there been hemorrhagic transformation?

4. If there was previous hemorrhage, has it increased?5. Has there been bleeding away from infarct?6. Is there hydrocephalus?7. Is there cerebral edema?8. Is there brain herniation?

What imaging examination is really the best?

Detection of blood products

NCCT has an overall sensitivity of approximately91% to 92% [50]. Decrease in accuracy with time islikely caused by evolution of blood density. FLAIRmay be positive as early as 23 minutes, and yieldssensitivity of 92% to 100% and specificity of 100%in small groups [50]. The T1-weighted MR imagingsequence is useful for identification of methemoglo-bin. Susceptibility (gradient recalled echo) sequencesare most useful for identification of hemosiderin butare also useful for acute blood products.

Fig. 11. A 47-year-old man with ataxia. Right PICA territory infarct (arrow) is illustrated on DWI (A) and onfollow-up noncontrast head CT (B).

59Modern Emergent Stroke Imaging

Detection of infarct

It is traditionally taught that MR imaging withDWI is superior [67–72] to NCCT and conven-tional MR imaging without DWI for the detectionof infarct. NCCT becomes more accurate approxi-mately 12 hours following presentation to theemergency department [72]. Data also suggestthat CTA and perfusion imaging provide improve-ment over NCCT, with CTA placing somewherebetween NCCT and MR imaging with DWI interms of statistical values [73]. The author’s resultsare similar to those in the literature [74–77].

Access and time of examinations

Availability of MR imaging scanners has been indi-cated as a limitation to the widespread use of con-ventional MR imaging and DWI for the diagnosisof acute stroke. CT scanners have a better penetra-tion in the community, and may be accessed moreeasily and faster. Moreover, the time needed toperform NCCT followed by CTA (approximately15 minutes) is comparable with an acute strokeprotocol MR image at the author’s facility. Theinclusion of an MR imaging angiogram, however,nearly doubles the time of the examination forMR imaging. The inclusion of NCCT with CTAremains necessary to exclude intracranial hemor-rhage, which might otherwise be overlooked be-cause of contrast enhancement.

Summary

Stroke remains a challenge for all physicians and animportant public health issue. Accurate and timelyneuroimaging may affect every facet of patient carewhere stroke is suspected. Access to advanced CT

and MR imaging techniques likely improves theability to detect infarct and brain at risk for infarc-tion but classical NCCT is also of use, especiallywhere other modalities are unavailable. Because ofthe greater penetration of CT scanners, speed ofexamination, and the absence of MR imaging safetyconsiderations with CT, CTA and CTP have becomefirst-line evaluations of the patient with symptomsof stroke. MR imaging, and especially DWI, havebecome excellent diagnostic tools and may ulti-mately serve as first-line examinations across theUnited States and internationally.

References

[1] Vo KD, Lin W, Lee JM. Evidence-based neuro-imaging in acute ischemic stroke. NeuroimagingClin N Am 2003;13:167–83.

[2] Broderick J, Brott T, Kothari R, et al. The GreaterCincinnati/Northern Kentucky Stroke Study: pre-liminary first-ever and total incidence rates ofstroke among blacks. Stroke 1998;29:415–21.

[3] Williams GR. Incidence and characteristics of to-tal stroke in the United States. BMC Neurol 2001;1:2.

[4] Grunwald I, Reith W. Non-traumatic neurologi-cal emergencies: imaging of cerebral ischemia.Eur Radiol 2002;12:1632–47.

[5] Albers GW, Amarenco P, Easton JD, et al.Antithrombotic and thrombolytic therapy forischemic stroke: the Seventh ACCP Conferenceon Antithrombotic and Thrombolytic Therapy.Chest 2004;126:483S–512S.

[6] Rother J. CT and MRI in the diagnosis of acutestroke and their role in thrombolysis. ThrombRes 2001;103(Suppl 1):S125–33.

[7] Burnette WC, Nesbit GM, Barnwell SL. Intra-arterial thrombolysis for acute stroke. Neuro-imaging Clin N Am 1999;9:491–508.

60 Mullins

[8] Jager HR. Diagnosis of stroke with advancedCT and MR imaging. Br Med Bull 2000;56:318–33.

[9] Tegos TJ, Kalodiki E, Sabetai MM, et al. Stroke:pathogenesis, investigations, and prognosis—Part II of III. Angiology 2000;51:885–94.

[10] Kidwell CS, Villablanca JP, Saver JL. Advances inneuroimaging of acute stroke. Curr AtherosclerRep 2000;2:126–35.

[11] Lev MH, Nichols SJ. Computed tomographicangiography and computed tomographic perfu-sion imaging of hyperacute stroke. Top MagnReson Imaging 2000;11:273–87.

[12] Burnette WC, Nesbit GM. Intra-arterial throm-bolysis for acute ischemic stroke. Eur Radiol2001;11:626–34.

[13] Lev MH, Segal AZ, Farkas J, et al. Utility ofperfusion-weighted CT imaging in acute middlecerebral artery stroke treated with intra-arterialthrombolysis: prediction of final infarct volumeand clinical outcome. Stroke 2001;32:2021–8.

[14] Lev MH, Farkas J, Rodriguez VR, et al. CTangiography in the rapid triage of patients withhyperacute stroke to intraarterial thrombolysis:accuracy in the detection of large vessel throm-bus. J Comput Assist Tomogr 2001;25:520–8.

[15] Warach S. New imaging strategies for patientselection for thrombolytic and neuroprotectivetherapies. Neurology 2001;57:S48–52.

[16] Laloux P. Intravenous rtPA thrombolysis inacute ischemic stroke. Acta Neurol Belg 2001;101:88–95.

[17] Sunshine JL, Bambakidis N, Tarr RW, et al. Bene-fits of perfusion MR imaging relative to diffu-sion MR imaging in the diagnosis and treatmentof hyperacute stroke. AJNR Am J Neuroradiol2001;22:915–21.

[18] McCullough LD, Beauchamp NB, Wityk R. Re-cent advances in the diagnosis and treatmentof stroke. Surv Ophthalmol 2001;45:317–30.

[19] Heiss WD, Forsting M, Diener HC. Imaging incerebrovascular disease. Curr Opin Neurol 2001;14:67–75.

[20] Keris V, Rudnicka S, Vorona V, et al. Combinedintraarterial/intravenous thrombolysis for acuteischemic stroke. AJNR Am J Neuroradiol 2001;22:352–8.

[21] Schaefer PW, Grant PE, Gonzalez RG. Diffusion-weighted MR imaging of the brain. Radiology2000;217:331–45.

[22] Schellinger PD, Jansen O, Fiebach JB, et al.Feasibility and practicality of MR imaging ofstroke in the management of hyperacute cere-bral ischemia. AJNR Am J Neuroradiol 2000;21:1184–9.

[23] Rosenwasser RH, Armonda RA. Diagnostic imag-ing for stroke. Clin Neurosurg 2000;46:237–60.

[24] Nakano S, Iseda T, Kawano H, et al. Correlationof early CT signs in the deep middle cerebral ar-tery territories with angiographically confirmedsite of arterial occlusion. AJNR Am J Neuroradiol2001;22:654–9.

[25] Moonis M, Fisher M. Imaging of acute stroke.Cerebrovasc Dis 2001;11:143–50.

[26] Lee BI, Nam HS, Heo JH, et al. Yonsei StrokeRegistry: analysis of 1,000 patients with acutecerebral infarctions. Cerebrovasc Dis 2001;12:145–51.

[27] Kilpatrick MM, Yonas H, Goldstein S, et al.CT-based assessment of acute stroke: CT, CTangiography, and xenon-enhanced CT cerebralblood flow. Stroke 2001;32:2543–9.

[28] Gleason S, Furie KL, Lev MH, et al. Potentialinfluence of acute CT on inpatient costs in pa-tients with ischemic stroke. Acad Radiol 2001;8:955–64.

[29] Sorensen AG, Buonanno FS, Gonzalez RG, et al.Hyperacute stroke: evaluation with combinedmultisection diffusion-weighted and hemody-namically weighted echo-planar MR imaging.Radiology 1996;199:391–401.

[30] Shen WC, Lee CC. Computerized tomographicangiography in the evaluation of cerebral infarc-tion. Chung Hua I Hsueh Tsa Chih (Taipei)1999;62:255–60.

[31] Graf J, Skutta B, Kuhn FP, et al. Computedtomographic angiography findings in 103 pa-tients following vascular events in the posteriorcirculation: potential and clinical relevance.J Neurol 2000;247:760–6.

[32] Lee KH, Cho SJ, Byun HS, et al. Triphasicperfusion computed tomography in acute mid-dle cerebral artery stroke: a correlation with an-giographic findings. Arch Neurol 2000;57:990–9.

[33] Hunter GJ, Hamberg LM, Ponzo JA, et al. As-sessment of cerebral perfusion and arterial anat-omy in hyperacute stroke with three-dimensionalfunctional CT: early clinical results [see com-ments]. AJNR Am J Neuroradiol 1998;19:29–37.

[34] Lieberman G, Abramson R, Volkan K, et al. Tu-tor versus computer: a prospective comparisonof interactive tutorial and computer-assistedinstruction in radiology education. Acad Radiol2002;9:40–9.

[35] Burdette JH, Ricci PE, Petitti N, et al. Cere-bral infarction: time course of signal intensitychanges on diffusion-weighted MR images. AJRAm J Roentgenol 1998;171:791–5.

[36] Fisher CM. A career in cerebrovascular disease:a personal account. Stroke 2001;32:2719–24.

[37] Mullins ME, Lev MH, Bove P, et al. Compari-son of image quality between conventional andlow-dose nonenhanced head CT. AJNR Am JNeuroradiol 2004;25:533–8.

[38] Mullins ME, Lev MH, Schellingerhout D, et al.Influence of availability of clinical history ondetection of early stroke using unenhanced CTand diffusion-weighted MR imaging. AJR Am JRoentgenol 2002;179:223–8.

[39] Tissue plasminogen activator for acute ischemicstroke. The National Institute of NeurologicalDisorders and Stroke rt-PA Stroke Study Group.N Engl J Med 1995;333:1581–7.

[40] Beauchamp Jr NJ, Barker PB, Wang PY, et al.

61Modern Emergent Stroke Imaging

Imaging of acute cerebral ischemia. Radiology1999;212:307–24.

[41] Furlan A, Higashida R, Wechsler L, et al. Intra-arterial prourokinase for acute ischemic stroke.The PROACT II study: a randomized controlledtrial. Prolyse in Acute Cerebral Thromboembo-lism. JAMA 1999;282:2003–11.

[42] Hacke W, Kaste M, Fieschi C, et al. Intrave-nous thrombolysis with recombinant tissue plas-minogen activator for acute hemispheric stroke.The European Cooperative Acute Stroke Study(ECASS). JAMA 1995;274:1017–25.

[43] Henson JW, Nogueira RG, Covarrubias DJ, et al.Gadolinium-enhanced CT angiography of thecircle of Willis and neck. AJNR Am J Neuroradiol2004;25:969–72.

[44] Mullins ME, Schaefer PW, Lev ML, et al. Intra-cranial hemorrhage complicating acute strokein a large emergency department series: Howcommon is it and is screening head CT neces-sary? In: Proceedings of the American Societyof Neuroradiology (ASNR) Annual Meeting. OakBrook (IL): ASNR; 2001. p. 321–2.

[45] Lev MH, Romero JM, Goodman DN, et al. Totalocclusion versus hairline residual lumen of theinternal carotid arteries: accuracy of single sec-tion helical CT angiography. AJNR Am J Neuro-radiol 2003;24:1123–9.

[46] Lam WW, So NM, Wong KS, et al. B0 imagesobtained from diffusion-weighted echo planarsequences for the detection of intracerebralbleeds. J Neuroimaging 2003;13:99–105.

[47] Lin DD, Filippi CG, Steever AB, et al. Detec-tion of intracranial hemorrhage: comparison be-tween gradient-echo images and b(0) imagesobtained from diffusion-weighted echo-planarsequences. AJNR Am J Neuroradiol 2001;22:1275–81.

[48] Jaillard A, Hommel M, Baird AE, et al. Signifi-cance of early CT signs in acute stroke: a CT scan-diffusion MRI study. Cerebrovasc Dis 2002;13:47–56.

[49] Ozsunar Y, Grant PE, Huisman TA, et al. Evo-lution of water diffusion and anisotropy in hy-peracute stroke: significant correlation betweenfractional anisotropy and T2. AJNR Am J Neu-roradiol 2004;25:699–705.

[50] Russell EJ, Angtuaco EJC, Elster AD, et al. SET42: Neuroradiology (second series) test andsyllabus. Reston (VA): The American College ofRadiology; 1998.

[51] Lev MH. CT versus MR for acute stroke imaging:is the ‘‘obvious’’ choice necessarily the correctone? AJNR Am J Neuroradiol 2003;24:1930–1.

[52] Mullins ME, Schwamm L, Maqsood M, et al.Assessment of the dense vessel sign on noncon-trast head CT I: evaluation of the hyperdensemiddle cerebral artery sign by CTA. In: Proceed-ings of the American Society of Neuroradiology(ASNR) annual meeting. Oak Brook (IL): ASNR;2004. p. 153–4.

[53] Kloska SP, Nabavi DG, Gaus C, et al. Acute stroke

assessment with CT: do we need multimodalevaluation? Radiology 2004;233:79–86.

[54] Hunter GJ, Silvennoinen HM, Hamberg LM,et al. Whole-brain CT perfusion measurementof perfused cerebral blood volume in acute is-chemic stroke: probability curve for regionalinfarction. Radiology 2003;227:725–30.

[55] Coutts SB, Lev MH, Eliasziw M, et al. ASPECTSon CTA source images versus unenhanced CT:added value in predicting final infarct extent andclinical outcome. Stroke 2004;35:2472–6.

[56] Lev MH, Farkas J, Gemmete JJ, et al. Acute stroke:improved nonenhanced CT detection–benefitsof soft-copy interpretation by using variablewindow width and center level settings. Radi-ology 1999;213:150–5.

[57] Naylor AR, Rothwell PM, Bell PR. Overview ofthe principal results and secondary analyses fromthe European and North American randomisedtrials of endarterectomy for symptomatic ca-rotid stenosis. Eur J Vasc Endovasc Surg 2003;26:115–29.

[58] Koenig M, Klotz E, Luka B, et al. Perfusion CT ofthe brain: diagnostic approach for early detectionof ischemic stroke. Radiology 1998;209:85–93.

[59] Schaefer PW, Hunter GJ, He J, et al. Predictingcerebral ischemic infarct volume with diffusionand perfusion MR imaging. AJNR Am J Neuro-radiol 2002;23:1785–94.

[60] Bonaffini N, Altieri M, Rocco A, et al. Functionalneuroimaging in acute stroke. Clin Exp Hyper-tens 2002;24:647–57.

[61] Wintermark M, Reichhart M, Cuisenaire O, et al.Comparison of admission perfusion computedtomography and qualitative diffusion- andperfusion-weighted magnetic resonance imag-ing in acute stroke patients. Stroke 2002;33:2025–31.

[62] Schaefer PW, Ozsunar Y, He J, et al. Assessingtissue viability with MR diffusion and perfu-sion imaging. AJNR Am J Neuroradiol 2003;24:436–43.

[63] Eastwood JD, Lev MH, Wintermark M, et al.Correlation of early dynamic CT perfusion imag-ing with whole-brain MR diffusion and perfusionimaging in acute hemispheric stroke. AJNR AmJ Neuroradiol 2003;24:1869–75.

[64] Sanelli PC, Lev MH, Eastwood JD, et al. The ef-fect of varying user-selected input parameters onquantitative values in CT perfusion maps. AcadRadiol 2004;11:1085–92.

[65] Goldstein LB, Jones MR, Matchar DB, et al.Improving the reliability of stroke subgroup clas-sification using the Trial of ORG 10172 in AcuteStroke Treatment (TOAST) criteria. Stroke 2001;32:1091–8.

[66] Kolominsky-Rabas PL, Weber M, Gefeller O,et al. Epidemiology of ischemic stroke subtypesaccording to TOAST criteria: incidence, recur-rence, and long-term survival in ischemic strokesubtypes: a population-based study. Stroke 2001;32:2735–40.

62 Mullins

[67] Bryan RN, Levy LM, Whitlow WD, et al. Diag-nosis of acute cerebral infarction: comparison ofCT and MR imaging [see comments]. AJNR AmJ Neuroradiol 1991;12:611–20.

[68] Barber PA, Darby DG, Desmond PM, et al. Iden-tification of major ischemic change: diffusion-weighted imaging versus computed tomography.Stroke 1999;30:2059–65.

[69] Lansberg MG, Albers GW, Beaulieu C, et al.Comparison of diffusion-weighted MRI and CTin acute stroke [see comments]. Neurology 2000;54:1557–61.

[70] Fiebach JB, Schellinger PD, Jansen O, et al. CTand diffusion-weighted MR imaging in random-ized order: diffusion-weighted imaging resultsin higher accuracy and lower interrater variabilityin the diagnosis of hyperacute ischemic stroke.Stroke 2002;33:2206–10.

[71] Lovblad KO. Diffusion-weighted MRI: back tothe future. Stroke 2002;33:2204–5.

[72] Mullins ME, Schaefer PW, Sorensen AG, et al. CTand conventional and diffusion-weighted MR

imaging in acute stroke: study in 691 patients atpresentation to the emergency department.Radiology 2002;224:353–60.

[73] Mullins ME, Cullen SP, Lev MH, et al. Acuteinfarct detection in a large emergency depart-ment series during the year 2000: accuracy of CTangiography (CTA). Radiology 2002;225(P):277.

[74] Beauchamp Jr NJ, Bryan RN. Acute cerebral is-chemic infarction: a pathophysiologic review andradiologic perspective. AJR Am J Roentgenol1998;171:73–84.

[75] Keir SL, Wardlaw JM. Systematic review of dif-fusion and perfusion imaging in acute ischemicstroke. Stroke 2000;31:2723–31.

[76] Arenillas JF, Rovira A, Molina CA, et al. Pre-diction of early neurological deterioration usingdiffusion- and perfusion-weighted imaging in hy-peracute middle cerebral artery ischemic stroke.Stroke 2002;33:2197–203.

[77] Tomandl BF, Klotz E, Handschu R, et al. Compre-hensive imaging of ischemic stroke with multi-section CT. Radiographics 2003;23:565–92.